Love me, Love me not…how VERN detects Love & Affection

Love…what a many splendored thing!

We have been hard at work researching the phenomena of communications, and specifically working on euphoric emotional states Love and Affection. Most of our clients have requested detection of the dysphoric emotional states-namely anger and sadness; but the detection isn’t a complete picture. As we state in our other blogs, as the emotion appears present in communications we analyze, and as our system matures more emotional states will “reveal” themselves. Which, is what happened with Love & Affection.

During our work with Michigan State University Federal Credit Union, we noticed some of this phenomenon. When clients would be gracious for their service, and also when they were satisfied with the interactions they’d express themselves with something similar to Love and Affection.

During our work with Boise State University’s GIMM Lab we noticed that there were other indications of an affection scale present.

And, in working with we coordinated research and development with their team, to conceptualize and to fine tune these detections in order to roll out a product for VERN AI and for Evrmore.

What is love?

We feel that love and affection are the same euphoric condition, on a scale of intensity. This intensity increases the more that the individual’s goals align with the subject. For example, a new car may fulfill the goal of appearing successful. Or, a hunky boyfriend may fulfill the goal of sexual attraction.

In this way, objects can (rightfully) be objects of affection. And, like in our own interpersonal communication, people and relationships factor into the VERN AI model.

There appears to be little to no incongruity between expectations and results, which makes Love & Affection a simpler emotional state and therefore has fewer emotives. You may recall from previous blogs, that emotives are sub-dimensions of the emotions themselves. These emotives combine to form for the receiver understanding of the sender’s intent.

Baby don’t hurt me…no more…

We have had some great results from the testing. We took hundreds of loving & affectionate phrases from crowdsourced and ranked sources, literature, poetry, music, and popular culture. Then we mixed in other salient phrases scored as sadness and anger from previous tests, and of course control statements from warranty contracts and technical writing sources.

The results?

Between a 92-96{b1ba36726a3bfcdc42af6e5ec24af305dbc6425c95dfb7052d7f2b4aabbf1a02} accuracy across the tests. That’s pretty good. We’d love for you to try it out yourself and let us know what results you got from it. To request a testing seat, please email us today.

We have also identified a bug that had been responsible for previous false-positives and have that ready for the next release (which will precede this detector…both coming soon).